用寄存器和对比对齐提升扩散模型的物体中心学习效果
Improved Object-Centric Diffusion Learning with Registers and Contrastive Alignment
- 引入寄存器槽吸收残差注意力，减少物体槽间干扰
- 通过对比对齐损失提升槽与图像内容的对应关系，提升6.1%发现准确率
- 方法轻量高效，适合复杂真实场景的物体识别与生成
Slot Attention 结合预训练扩散模型在物体中心学习中展现潜力，但存在槽位纠缠和槽与图像内容对齐弱的问题。本文提出对比物体中心扩散对齐（CODA），通过两种方式改进：(i) 使用寄存器槽吸收残差注意力，降低物体槽间的干扰；(ii) 引入对比对齐损失，显式增强槽与图像内容的对应关系。该训练目标可作为最大化槽与输入之间互信息的可计算代理，提升槽表示质量。在合成数据集（MOVi-C/E）和真实数据集（VOC、COCO）上，CODA 在物体发现（如 COCO 上 +6.1% FG-ARI）、属性预测和组合图像生成方面均优于强基线。寄存器槽带来极小计算开销，保持方法高效可扩展。结果表明，CODA 可作为复杂真实场景下鲁棒物体中心学习的有效框架。代码与预训练模型见 https://github.com/sony/coda。
原文摘要 · Abstract (English)
Slot Attention (SA) with pretrained diffusion models has recently shown promise for object-centric learning (OCL), but suffers from slot entanglement and weak alignment between object slots and image content. We propose Contrastive Object-centric Diffusion Alignment (CODA), a simple extension that (i) employs register slots to absorb residual attention and reduce interference between object slots, and (ii) applies a contrastive alignment loss to explicitly encourage slot-image correspondence. The resulting training objective serves as a tractable surrogate for maximizing mutual information (MI) between slots and inputs, strengthening slot representation quality. On both synthetic (MOVi-C/E) and real-world datasets (VOC, COCO), CODA improves object discovery (e.g., +6.1% FG-ARI on COCO), property prediction, and compositional image generation over strong baselines. Register slots add negligible overhead, keeping CODA efficient and scalable. These results indicate potential applications of CODA as an effective framework for robust OCL in complex, real-world scenes. Code and pretrained models are available at https://github.com/sony/coda.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。